Skip to content
DraconiticResearch
Search Sources API
203,203 studies31 sourcesUpdated 8 min ago
Peer-reviewedCC BY-NC 4.0La Medicina del Lavoro

Supporting Ergonomic Risk Assessment: An AI-Driven Approach for Action Recognition Using Standard Video Recordings

  • Beatrice Doro1
  • Francesco Favro1
  • Francesca Zen1
  • Valentina Bullo1,2
  • Stefano Mattioli3,4
  • Nicolò Navarin5,6
  • Stefano Gobbo1,2
  • Marco Bergamin1,2
6 affiliations
  1. 1Department of Medicine, University of Padua, Padua, Italy
  2. 2Gymhub S.r.l., Spin-off of the University of Padua, Padua, Italy
  3. 3Division of Occupational Medicine, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Italy
  4. 4Department of Medical and Surgical Sciences, University of Bologna, Italy
  5. 5Department of Mathematics, University of Padua, Padua, Italy
  6. 6Human Inspired Technology Research Centre, University of Padua, Padua, Italy
Read full textPDFDOICite

Abstract

Background:: While AI has been widely adopted in medical, social, and industrial fields, its use in ergonomic risk assessment of occupational movements is limited. Existing automated activity recognition methods often rely on controlled conditions or extra sensors, reducing their practicality. This study aims to develop an AI-driven tool to digitize the Occupational Repetitive Actions (OCRA) risk assessment by automatically identifying occupational actions from standard video recordings in real manufacturing settings. The goal is to help ergonomists conduct risk assessments more broadly and support workers’ health.

Methods:: Video data were collected from 37 workstations across multiple manufacturing facilities, comprising 869 action clips representing 12 distinct occupational activities. Two-dimensional skeletal pose extraction was performed using MMAction2 to anonymize workers and reduce environmental confounding factors. Graph Convolutional Network architectures (ST-GCN and Two-Stream Adaptive GCN) were trained and validated using stratified dataset partitioning (67.3% training, 24.7% validation, 8.0% test).

Results:: The proposed model achieved approximately 60% classification accuracy, substantially outperforming the frequency-based (12%) and duration-based (24%) baselines. The performance evaluation used multiple metrics, including Mean Average Precision, F1-score, Jaccard/IoU, Dice-Sørensen coefficient, and Hausdorff distance.

Conclusions:: This work demonstrates the technical feasibility of AI-driven, markerless motion capture systems for digitalizing musculoskeletal disorder risk assessment in real-world occupational settings. The principal innovation is the successful implementation of action recognition from standard video recordings captured in uncontrolled manufacturing environments, which transcends spatial constraints and eliminates the need for specialized sensing equipment. Despite current limitations in dataset size and classification accuracy, this study lays a foundation for developing deployable platforms to support real-time ergonomic assessment and proactive prevention of work-related musculoskeletal disorders.

Related

  1. Comparing a Dedicated Breast Spectral CT Protocol With Standard MonoE40 Reconstruction for Breast Cancer Characterization and Opportunistic Assessment
    Ashton Padowitz, Zane Sherif, Kirralee Sherif +1
    CureusPeer-reviewedCC BYFull textMedical imaging & radiologyOncologyEurope PMCPMC
  2. Ergonomic interventions for treating work‐related complaints of the arm, neck or shoulder in adults
    Poonam Mehta, Peter W Stubbs, Sietske J Tamminga +4
    The Cochrane Database of Systematic ReviewsPeer-reviewedPublisher termsFull textPharmacology & toxicologyPMC
  3. Intelligent surgical workflow recognition-based skill assessment for minimally invasive submandibular gland resection
    Zhongkai Ma, Yufei Hua, Lin Que +9
    NPJ Digital MedicinePeer-reviewedCC BY-NC-ND 4.0Full textSurgery & anaesthesiaPMC
  4. Diagnostic pathways for gastroesophageal reflux cough: From empirical recognition to MII-pH-based multidimensional assessment
    Chang Liu, Qun Wang, Lei Pan +6
    Therapeutic Advances in Respiratory DiseasePeer-reviewedCC BY-NC 4.0Full textGeneral medicinePMC
  5. High-flow nasal cannula versus standard oxygen therapy for adult acute hypoxemic respiratory failure: a systematic review and meta-analysis with trial sequential analysis and GRADE assessment
    Liangdong Zhu, Xun Yang, Jing Tan +1
    BMC Pulmonary MedicinePeer-reviewedCC BYFull textGeneral medicineDOAJ

Cite

BibTeXRIS
BibTeX
@article{doro2026supporting,
  title = {{Supporting Ergonomic Risk Assessment: An AI-Driven Approach for Action Recognition Using Standard Video Recordings}},
  author = {Beatrice Doro and Francesco Favro and Francesca Zen and Valentina Bullo and Stefano Mattioli and Nicolò Navarin and Stefano Gobbo and Marco Bergamin},
  journal = {La Medicina del Lavoro},
  year = {2026},
  month = oct,
  doi = {10.23749/mdl.2026.18921},
  url = {https://doi.org/10.23749/mdl.2026.18921},
  note = {CC BY-NC 4.0},
}
RIS
TY  - JOUR
TI  - Supporting Ergonomic Risk Assessment: An AI-Driven Approach for Action Recognition Using Standard Video Recordings
AU  - Beatrice Doro
AU  - Francesco Favro
AU  - Francesca Zen
AU  - Valentina Bullo
AU  - Stefano Mattioli
AU  - Nicolò Navarin
AU  - Stefano Gobbo
AU  - Marco Bergamin
PY  - 2026
DA  - 2026/10/23
JO  - La Medicina del Lavoro
DO  - 10.23749/mdl.2026.18921
UR  - https://doi.org/10.23749/mdl.2026.18921
AB  - Background:: While AI has been widely adopted in medical, social, and industrial fields, its use in ergonomic risk assessment of occupational movements is limited. Existing automated activity recognition methods often rely on controlled conditions or extra sensors, reducing their practicality. This study aims to develop an AI-driven tool to digitize the Occupational Repetitive Actions (OCRA) risk assessment by automatically identifying occupational actions from standard video recordings in real manufacturing settings. The goal is to help ergonomists conduct risk assessments more broadly and support workers’ health.  Methods:: Video data were collected from 37 workstations across multiple manufacturing facilities, comprising 869 action clips representing 12 distinct occupational activities. Two-dimensional skeletal pose extraction was performed using MMAction2 to anonymize workers and reduce environmental confounding factors. Graph Convolutional Network architectures (ST-GCN and Two-Stream Adaptive GCN) were trained and validated using stratified dataset partitioning (67.3% training, 24.7% validation, 8.0% test).  Results:: The proposed model achieved approximately 60% classification accuracy, substantially outperforming the frequency-based (12%) and duration-based (24%) baselines. The performance evaluation used multiple metrics, including Mean Average Precision, F1-score, Jaccard/IoU, Dice-Sørensen coefficient, and Hausdorff distance.  Conclusions:: This work demonstrates the technical feasibility of AI-driven, markerless motion capture systems for digitalizing musculoskeletal disorder risk assessment in real-world occupational settings. The principal innovation is the successful implementation of action recognition from standard video recordings captured in uncontrolled manufacturing environments, which transcends spatial constraints and eliminates the need for specialized sensing equipment. Despite current limitations in dataset size and classification accuracy, this study lays a foundation for developing deployable platforms to support real-time ergonomic assessment and proactive prevention of work-related musculoskeletal disorders.
SN  - 0025-7818
SN  - 2532-1080
ER  -